Analogue and Numerical Models of Earthquake Rupture

Abstract

Earthquakes represent one of the most important natural risks facing human populations in urban areas. Understanding the processes at the origin of these destructive events requires seismological observations, but also the use of laboratory analogues and numerical models for earthquake rupture. They allow for controlled conditions under which we can investigate the relative importance of different physical quantities involved in the system. The main points investigated in this thesis are the influence of loading rate on the nucleation of earthquakes, and the evolution of friction during dynamic ruptures. I conduct photoelastic experiments using polycarbonate plates, but also direct-shear experiments of precut granite blocks in a pressure vessel. I use finite-difference numerical models to reproduce and understand the dynamic laboratory ruptures, and I developed static finite element codes in order to reproduce the loading conditions induced by the experimental setup. The main results are that under certain conditions,
increasing the loading rate makes the nucleation length shrink, and affects the nucleation position, which in this case is consistently situated on high coulomb stress areas. This is not necessarily the case for low loading rates. The shrinking of nucleation length may explain partly why some asperities in subduction zones can behave seismically or aseismically depending on the local tectonic loading
velocity. Finally, I propose a method to estimate the dependence of friction on slip and slip velocity from strain gauge data during friction experiments. When conducted under realistic pressure conditions, this can provide useful constitutive laws to implement in numerical models simulating earthquakes. Eventually, the results presented in this thesis can be used in order to improve rupture scenarios, and short-term earthquake forecast.